Simulation Study Assessing the Impacts of Automated Vehicles on Parking Demand in San Francisco’s City Center Saw a 72 Percent Reduction in CO2 Emissions Compared to a No-Automated Vehicles Scenario.

Researchers Assessed Impacts of Personal Automated Vehicle Parking in San Francisco's Downtown Central Business District Using Traffic Simulation Model with Local Travel Activity Data.

Date Posted

The Impacts of Automated Vehicles on Center City Parking Demand

Summary Information

Automated Vehicles (AV) have the capability to drop-off (DO) and pick-up (PU) passengers as needed which comes in handy in areas where parking is scarce or available only at high costs. This non-reliance of AVs on parking spots could in turn present potential opportunities for redevelopment that could improve the livability of cities, such as enhanced sidewalk space for pedestrian and bicycle travel. However, reduced demand for parking would be accompanied by increased demand for curbside DO/PU space with related movements to enter and exit the flow of traffic. Given this view, this study used a microscopic traffic simulation model with local travel activity data to simulate personal AV parking scenarios in San Francisco's downtown Central Business District (CBD). 


An open-source microscopic traffic simulation software was used to model the study area. The simulation model used the local travel activity data from the San Francisco Bay Area’s activity-based travel demand model.  It was assumed in the simulation model that both on-street parking and DO/PU events would take place at the on-street at the curb only (i.e., no double parking). The study selected individual daily activity tours with at least one vehicle stop from the travel demand model. The study also converted transit trips to AV trips for the purpose of the simulation. The San Francisco Parking Census data were used to estimate the parking supply. Four scenarios were considered in the study:

  • Scenario 1: Simulated 100 percent parking, incrementally increased DO/PU (drop-offs/pick-ups) trips in 10 percent steps until reaching 100 percent.
  • Scenario 2: Incrementally converted on-street parking to DO/PU spaces, assuming 80 percent parking demand and 20 percent DO/PU, with constant travel activity.
  • Scenario 3: Used 50 percent DO/PU traffic from Scenario 1, without dedicated DO/PU spaces.
  • Scenario 4: Simulated a 30 percent traffic reduction due to auto pricing policies, with varying on-street parking and DO/PU shares.

Travel patterns were held constant in all scenarios, meaning changes in congestion did not affect the demand for travel, parking, or DOs/PUs in the CBD. The study assumed the use of personal automated vehicles (AVs), not shared AVs.



  • Scenario 1 results showed that speeds went as low as one meter/second (or about two mph) during the mid-day period in the zero percent scenario (all vehicles park) and increased rapidly to free-flow speeds (eight meter/second or 18 mph), when DO/PU trips substitute for 40 percent to 50 percent of parking trips.
  • Scenario 4 results indicated a 72 percent reduction in CO2 emissions compared to a no-AV scenario where 100% of trips park.
Results Type
Deployment Locations